Diary Entries in English

Recent diary entries

Posted by Renanuma on 21 November 2026 in English.

Hello, my name is Numa Garen-ere, Im a student of federal university, Otuoke. Department of environmental management and toxicology. I was invited for state of the map Nigeria Uyo 2026 conference in Uyo. It was a great experience as it was my first time attending. We learned a great deal about mapping and discoveries, I also had a great time after the conference, we went for tour and it was amazing to see the culture of Uyo around the environment. SOTM was a great experience for me. Thank you all for this opportunity given to us.

Posted by skfd on 20 November 2026 in English.

StreetComplete is on iPhone now, and I’m question so very much on my long walks. Not enjoying fall landscapes no, not noticing quirky villager buildings – just glancing around and mostly looking at the screen tapping away… I love StreetComplete too much, so fun and useful.

GoMap!! is also great so I even survey the sidewalks and bike parking and what not in it. Same problem – too much screentime.

I also feel uneasy capturing all the easy obvious things How would I involve new people into hobby, if the only thing left to survey would be roof shapes and weather restaurant sells kosher?

I think I should leave fun easy quests and tasks for new mappers, and focus on imports, community organizing, and tool making.

Location: Hanlon Creek Neighbourhood Group, Guelph, Southwestern Ontario, Ontario, Canada
Posted by Jack_Regan on 20 November 2026 in English.

A while back I noticed someone had mapped a farmyard i.e. the boundary of the farmyard only, but there was none of the farm buildings mapped within it. So I added them.

Then I found another, same story; then another, there was a lot of them! Farmyards with zero buildings.

This is a job for Overpass Turbo [OT], I thought. My first attempt was to use the “around” command or filter or whatever its proper name is. Find farmyards with no buildings within a certain distance. I’m certainly no Overpass Turbo expert but I can write my own queries as long as they’re not too complex.

My method didn’t work very well, so I asked ChatGPT for some help. It soon told me I was using the wrong method and with a little bit of effort we eventually had a working query.

A couple of years ago chatbots/LLMs were absolutely hopeless at writing OT queries, often outputting complete nonsense. Now? They can write and debug complex queries far better than I can.

I live in Perth and Kinross, so the query was built around the relevant relation (ID: 1915429) for the P and K council area (it’s very large).

Overpass Turbo found somewhere in the region of 1500+ farmyards in the area. Roughly one third had no buildings (as at 02 October 2026) so a lot of mapping to do.

I saved the results of the query and imported it into JOSM as a new layer so I wouldn’t have to run the query again.

I’ve been working through them, trying to do at least 4 or 5 every day.

The same query could easily be altered to find residential landuse with no buildings, or blank construction areas which now have buildings waiting to be mapped.

Obviously, the relation ID can also be switched to work another area.

Future tasks

Find the farms with no name, and research the name (old OS map?)

Repeat for landuse=residential with no buildings mapped (a quick checked returned around 300?! Seems high?

Posted by Peter Elderson on 19 November 2026 in English.

(I’ll keep the roundabouts round for this article; other shapes are not that different where it comes to mapping.)

Structure

Usually there is a central circle, the central natural/landuse/leisure, surrounded by a paved ring, surrounded by a wedgeshaped circle of kerbstones: the apron (aka truck apron).

Central natural/leisure/landuse

The days of plain grass roundabout islands are over, in Nederland. You can still find them, but most centre islands have bushes, art, trees, shrubberies, hedges, flowerbeds, statues and monuments, even hills, paths, walls and landscape examples. Many are used for commercial purposes, e.g. as garden showcases.

My take on that is to map it as what it’s made to look like. If it is made to look like a beach, garden, wood or park, I tag it as a round beach, garden, wood or park, usually with access=no (there are exceptions even to that!).

The apron ring

The apron ring is most often paved with stones, but I have seen many other materials. My take on this is: I don’t map the apron ring itself, only the apron kerb ring.

The apron kerb

Apron kerbs are designed as easily traversable barriers, but the further you traverse them, the more unpleasant it gets. Regular cars will usually not traverse the apron kerb, even though it would not harm the car. I map the kerb by drawing the outside kerb circle and tag it as barrier=kerb, + kerb=apron (ATYL).

The whole island

The central island is the whole area within the apron kerb way, including the central natural. I tag it as area:highway=central_island (YOLO, I mean, ATYL). I could put the tag on the apron kerb way, but that mixes features as wel as data types. Luckily, there is an established method to use tagged way elements a outlines of an area: the multipolygon. In this case the MP is very simple: only one outer, no inner. The apron kerb way is the outer, and the area:highway=central_island tag goes on the MP area.

… See full entry

Good maps grow through careful observation, clear sources, and respectful review. Before changing a feature, it helps to check the existing data, explain the reason in the changeset, and confirm that the published result reflects the intended edit.

For boundary-related work, I think it is especially useful to keep official reference maps clearly attributed and separate from OpenStreetMap’s community-maintained data. A detected change should lead to an evidence-based review of the object history and sources, rather than assumptions about the contributor’s intentions.

This general project note is shared by the operator across accounts in the same mapping workspace. It does not announce a map edit or claim a community consensus.

Good maps grow through careful observation, clear sources, and respectful review. Before changing a feature, it helps to check the existing data, explain the reason in the changeset, and confirm that the published result reflects the intended edit.

For boundary-related work, I think it is especially useful to keep official reference maps clearly attributed and separate from OpenStreetMap’s community-maintained data. A detected change should lead to an evidence-based review of the object history and sources, rather than assumptions about the contributor’s intentions.

This general project note is shared by the operator across accounts in the same mapping workspace. It does not announce a map edit or claim a community consensus.

Posted by Peter Elderson on 18 November 2026 in English.

When I joined the OSM-community, landcover was being proposed. The idea ws to get non-landuse landuse tags and non-natural natural tags out of the landuse and natural keys. I especially dislike landuse=grass because grass is not landuse. If grass is used for something, it is meadow or farmland or e.g. decoration, separation and the like. So the landcover seemed a goed idea to me.

Since then, I’ve changed my view. Only two values make sense: grass and trees. Trees already has two alternatives: natural=wood and landuse=forest. At the data user side, both just mean an areas of mainly trees. Landcover=trees only adds a third option to tag the same thing. Grass has three estrablished tags: landuse=grass, landuse=meadow and natural=grassland. These are more distinctive than the tree tags, but that is mainly at the mapper’s side; they think they make a distinction, but on the data user side it’s all just land covered with grass.

All the other proposed values are covered by the appropriate values in the natural key, and there is nothing wrong with that: objects in the natural key are objects which grow, flow or form by nature: life forms, and geological forms, in contrast to man_made objects. When man mimics natural objects they are still categorized as naturals: beach, water bodies, tree rows. Workable.

And, trees and grass also have appropriate natural tags: natural=wood and natural=grassland. This makes a move to the landcover key unnecessary. Combined with the fact that currenty the landcover tags are usually added to one of the established tags for grass and trees, which means that the full transition will never occur, I now opt for dropping the landcover key and using established naturals instead.

… See full entry

Posted by Harry Wood on 17 November 2026 in English.

The next London OSM meet-up will be December 9th at the Monkey Puzzle.

Last time we were at “The Volunteer” pub! Obviously with a name like that, I hoped this would be a good pub for us to have our gathering of volunteer mappers, and to show solidarity with the unpaid bar staff. But it turns out this pub was pretty noisy and busy, and what’s more, I suspect the staff were actually being paid :-) I booked a table for 8 but got a table with four places. I suppose the idea was that it’s big enough for 8 people to drink around, which we put to the test, and then about 12 people arrived, and then we ordered food. So it was a bit of squeeze, but fun!

OSMLondon at the volunteer (not quite during peak crowdedness)

User spiregrain was there (impressive editing run!). He’s from Statford, so we talked about mapping there, and we were remembering having a massive Westfield shopping mall mapping party there. Looking back at the old diary entry now, I see this was 2011. I’d forgotten we organised to go there shortly after the mall opened.

… See full entry

Location: East Marylebone, Mayfair, City of Westminster, Greater London, England, W1T 3PP, United Kingdom

I run Aloe Energy, a small not-for-profit initiative working on energy access in Sub-Saharan Africa. We published a free interactive map of the Southern African Power Pool grid, and OpenStreetMap is its backbone, so I wanted to share how it was built and where OSM could use help.

What we did. We took every OSM power line of 66 kV and above across the 12 SAPP countries (country extracts from November 2026). Where OSM had no line but an older public map did, we filled the gap from TANESCO (2015) and World Bank (2017) grid maps, but only for lines recorded as existing, with a known voltage, and more than about 3 km from any OSM line. A “Source and age” view on the map colors each line by where it comes from, so you can see exactly what is OSM and what is gap-fill.

Where OSM is thin. Coverage is weakest in Tanzania, DR Congo and Angola. Those are the places where the map still leans on maps that are close to ten years old, and where new mapping would change the picture most. Substations named in national plans are often missing too.

The map: https://aloeenergy.org/research/sapp-grid. There are also maps for the Eastern Africa Power Pool and Madagascar built the same way.

OSM data is credited as “OpenStreetMap contributors (ODbL)”, and the OSM-derived line data is available on request under the ODbL. The project register and plant list we compiled are free to download under CC BY 4.0 at https://aloeenergy.org/data. If you map power in the region and spot something wrong, I would be glad to hear about it and point people to fix it in OSM.

Image: Students and faculty coming together to build the Native Mappers community at HNBGU, Uttarakhand.

Native Mappers First Meeting at HNBGU

22 August 2026 Department of Geography, Chauras Campus, Hemvati Nandan Bahuguna Garhwal University, Uttarakhand, India

A new student-led mapping initiative, Native Mappers, is taking shape at Hemvati Nandan Bahuguna Garhwal University (HNBGU), Uttarakhand. The initiative aims to bring university students closer to OpenStreetMap (OSM), collaborative mapping and the wider open mapping community.

The first meeting at the Department of Geography, Chauras Campus, introduced students to the idea of building a university-based mapping community where they can learn practical skills, contribute geographic information and work together on local mapping projects.

Connecting Local Mapping with Global Opportunities

… See full entry

Location: Devprayag, Tehri Garhwal, Uttarakhand, India

History

In 2010, a paper map by 3dShapes was digitized and imported, to map landuse in Nederland. The import project was well conducted and properly documented in the OSM-wiki. This was a major step in filling in the space on OSM.

Some problems remained after import, to be dealt with later. For instance, where trees had been planted in multiple rows, it was imported as an orchard. As a result, our motorways were suddenly lined with orchards, on OSM-based maps. This has been fixed by organized mappings. Another example: the paper map had a tile grid of thin black lines. These lines were seen as cut lines, resulting in unnecessary and awkward splits of the imported areas. The import procedure did detect this problem, but couldn’t automate a fix, so this was left to the future mappers. It took 15 years before a crazy mapper got rid of this in a solo project lasting 2 months. Guess who?

The issue

Not much of a problem, more of an annoyance, really: All polygons were tagged with “source=3dShapes”. A significant part of the polygons still carry this source tag. Many polygons have not been touched since import; many others have been edited but kept the original source tags; and many have been significantly altered or replaced thereby losing the original source tag. As a result, the tag is now generally meaningless. It only tells you that the former 3dShapes donated their map, but it tells you nothing about the current size, shape, object type, and tagging. These can be from unaltered to completely different.

The current consensus is that source tags should not be attached to objects, but to changesets. Why not just mass remove this outdated source tag?

Mass remove the 3dShapes source tag?

… See full entry

Posted by nickvet419 on 15 November 2026 in English.

osm.org/changeset/105681 One of my Changesets from 2008 when I proposed service=parking_aisle osm.wiki/Proposal:Service%3Dparking_aisle

Changesets were implemented in April 2009 API v0.6. My first retroactive changeset recorded was 1 Jun 2008  osm.org/changeset/11289 where I was mapping out Nevis Island where one of my friends had gone to med school. Anything I did prior wasn’t recorded, but I started mapping Chicago area roads and highways November 08, 2007.

Posted by CodentGIS on 15 November 2026 in English.

It’s been two months since I started to edit the Ilagan River again. While I had done so in the past, way before these two months, this period turned out to be a great adventure that I’m still hiking.

It was a time after Project CaCau. When I put it to an end, I felt not only disappointed, but ambitious. Why exactly? Well, I don’t know, but the roaring engine I had used to keep me going through that project was still on, and I couldn’t just stop it. So, I had to redirect it somewhere.

For a bit, I had just redirected it to OSM. I experimented with multiple things and tried to pick up my old stuff. But one in particular caught my attention. A while ago, I was editing the river area of the Ilagan River. I never picked it up again until then. Despite that, it just stuck. Starting in the 15th of September, I got into the flow of the edits, and I never stopped.

I didn’t just do the main river either. I had also started doing its tributaries because my satisfaction didn’t get put with just that. Then, suddenly, the Bintacan River got finished, then the Nanaguan, and then I was suddenly speeding through the headwaters of the Abuan. I was now pushing 2K-3K edits a day, with the little free time I had.

My satisfaction didn’t stop; I had also started mapping the landslide areas. Now, at that point, I was satisfied. I had forgotten Project CaCau and its shadow that burdened me. Then, I realized: why not make this run official? Make it a project? So, I did.

Enter, Project KarayanIlagan. It meant “Ilagan River” in Ilocano: Karayan as River, and Ilagan as itself. Nothing changed in my workflow; just that I finally had a name to keep it going and back it up, putting hashtag comments for it as #KarayanIlagan (and later on with #ChasingWaterways for my fellow mappers’ initiative), and to give it a history and recognition.

… See full entry

Location: Bangag, Ilagan, Isabela, Cagayan Valley, Philippines
Posted by fghj753 on 15 November 2026 in English.

I have custom branch of EveryDoor since July. Namely because for recycling containers i need tagging preset with custom field types not supported by iD schema and for ED plugin engine which has a annoying bug regarding NeedsInfo selectors.
Opening and searching presets was occasionally super slow and past few nights ive tried to fix annoyances. Turned out the performance fix was trivial as i had compiled app with wrong parameter which caused app to not be able to compile something related to presets lookup.

Anyways, 2 nights later, achievements include:

  • saving downloaded map data at least 20x faster (by using already existing db index which existing query disregarded)
  • zoom no longer jams at zoom level 15. Technical reason is that ED has internally 2 map views for different zoom ranges and ones min and other’s max zoom was lvl 15.
  • amenities list scrolling is now vertical instead horizontal (which i thought was unintuitive and waste of space)
  • tiles are loaded while zooming, in current version, after zooming tiles are downloaded only on map drag event. Fixed by updating map library.
  • i discovered ED already has special overview map of downloaded osm data (dark-green rectangles at Menu -> Data Management -> Download tiles), but they are rather useless as they denote simply areas that was downloaded at some point. Merged rectangles into rasterized layer to show up to date and outdated areas as one.
  • haven’t been able to test, but cached map tiles should cleared after 1 month to prevent out of date background layer. I have previously created multiple false positive notes about missing road or building only to find my map layer was outdated
  • dragging to add new element while moving and tracking gps, won’t cause unexpected jumps and zoom changes as user’s location changes.
  • few more changes related to map interactions i forgot

… See full entry

Posted by erickdeoliveiraleal on 14 November 2026 in English. Last updated on 15 November 2026.

If you map rural areas, you know the problem: Bing or Esri imagery is old, cloudy, or simply doesn’t show whether a road is paved or dirt. Sentinel-2 solves a good part of that. It is free satellite imagery, refreshed every few days, with 10 m resolution and infrared bands the human eye can’t see.

In the false color composite (bands B8, B4 and B3), vegetation turns red and asphalt turns dark and high-contrast. Paved roads crossing scrubland or farmland really stand out.

Sentinel-2 false color in JOSM: the paved road shows up as a dark, sharp line and vegetation is red

Sentinel-2 false color in JOSM, rural Ceará, Brazil. Contains modified Copernicus Sentinel data (2026). Image: Wikimedia Commons, CC BY-SA 4.0.

This tutorial shows how to add this layer to JOSM using the free Copernicus Data Space Ecosystem service from the European Union.

Important limitation: 10 m per pixel works for highways, wide rural roads, water, vegetation, land use and burned areas. It does not work for buildings or narrow urban streets.


1. Create a Copernicus Data Space account

… See full entry

Posted by rphyrin on 13 November 2026 in English.

“The idea is pretty cool. It downloads regional data from Geofabrik directly in the browser and formats the data into Parquet format, then lets you ask the AI ​​to query it using DuckDB.”

“This project looks cool, even though I’m still struggling to understand what use case it is trying to solve. (What does “routable road network” even mean?) (But well.. Perhaps this is partly due to my own bias, as I’ve never worked on any OSM-based routing or simulation projects)”

“I see data export to SUMO and MATSim, and LinkedIn mentions the possibility of using NetworkX. Judging by the profile, the author is a researcher with a focus on graph algorithms.”


“Export to PMTiles, MBTiles, GeoPackage, GeoJSON, kepler.gl, MATSim or SUMO”

… See full entry